Explore the vast range of titles available.

Specialized classes of proteins, working together in a tightly orchestrated manner, induce and maintain highly curved cellular and organelles membrane morphology. Due to the various experimental constraints, including the resolution limits of imaging techniques, it is non-trivial to accurately elucidate interactions among the various components involved in membrane deformation. The spatial and temporal scales of the systems also make it formidable to investigate them using simulations with molecular details. Interestingly, mechanics-based mesoscopic models have been used with great success in recapitulating the membrane deformations observed in experiments. In this review, we collate together and discuss the various mechanics-based mesoscopic models for protein-mediated membrane deformation studies. In particular, we provide an elaborate description of a mesoscopic model where the membrane is modeled as a triangulated sheet and proteins are represented as either nematics or filaments. This representation allows us to explore the various aspects of protein–protein and protein–membrane interactions as well as examine the underlying mechanistic pathways for emergent behavior such as curvature-mediated protein localization and membrane deformation. We also put forward current efforts in the field towards back-mapping these mesoscopic models to finer-grained particle-based models—a framework that could be used to explore how molecular interactions propagate to physical scales and vice-versa. We end the review with an integrative-modeling-based road map where experimental imaging micrograph and biochemical data are combined with mesoscopic and molecular simulations methods in a theoretically consistent manner to faithfully recapitulate the multiple length and time scales in the membrane remodeling processes.

Advances in coarse-grained molecular dynamics (CGMD) simulations have extended the use of computational studies on biological macromolecules and their complexes, as well as the interactions of membrane protein and lipid complexes at a reduced level of representation, allowing longer and larger molecular dynamics simulations. Here, we present a computational platform dedicated to the preparation, running, and analysis of CGMD simulations. The platform is built on a completely revisited version of our Martini coarsE gRained MembrAne proteIn Dynamics (MERMAID) web server, and it integrates this with other three dedicated services. In its current version, the platform expands the existing implementation of the Martini force field for membrane proteins to also allow the simulation of soluble proteins using the Martini and the SIRAH force fields. Moreover, it offers an automated protocol for carrying out the backmapping of the coarse-grained description of the system into an atomistic one.

A tight and consistent link between resolutions is crucial to further expand the impact of multiscale modeling for complex materials. We herein tackle the generation of condensed molecular structures as a refinement-backmapping-of a coarse-grained (CG) structure. Traditional schemes start from a rough coarse-to-fine mapping and perform further energy minimization and molecular dynamics simulations to equilibrate the system. In this study we introduce DeepBackmap: A deep neural network based approach to directly predict equilibrated molecular structures for condensed-phase systems. We use generative adversarial networks to learn the Boltzmann distribution from training data and realize reverse mapping by using the CG structure as a conditional input. We apply our method to a challenging condensed-phase polymeric system. We observe that the model trained in a melt has remarkable transferability to the crystalline phase. The combination of data-driven and physics-based aspects of our architecture help reach temperature transferability with only limited training data.

Coarse-grained (CG) model has been a powerful tool in bridging the gap between theoretical studies and experimental phenomena in biological computing field. The reconstruction from a CG model to an atomic-detail structure is especially important in CG studies of biological systems. In this work, a rigid-fragment- and local-frame-based (RF-LF) backmapping method was proposed to achieve reverse mapping from CG models to atomic-level structures. The initial atomic-level structures were further refined to yield the final backmapping ones. With the popular Martini force field, the performance of the RF-LF method was extensively examined in the CG → AA (CG to AA) backmapping of protein/DNA/RNA systems. Besides, the RF-LF method was also extended to the backmapping of the TMFF model. Numerical results illustrate that the RF-LF backmapping method is generic and parameter-free and can provide a promising way to tackle atomic-level studies in CG models.

Background Compared to classical genotyping, targeted next-generation sequencing ( t NGS) can be custom-designed to interrogate entire genomic regions of interest, in order to detect novel as well as known variants. To bring down the per-sample cost, one approach is to pool barcoded NGS libraries before sample enrichment. Still, we lack a complete understanding of how this multiplexed t NGS approach and the varying performance of the ever-evolving analytical tools can affect the quality of variant discovery. Therefore, we evaluated the impact of different software tools and analytical approaches on the discovery of single nucleotide polymorphisms (SNPs) in multiplexed t NGS data. To generate our own test model, we combined a sequence capture method with NGS in three experimental stages of increasing complexity ( E. coli genes, multiplexed E. coli , and multiplexed HapMap BRCA1/2 regions). Results We successfully enriched barcoded NGS libraries instead of genomic DNA, achieving reproducible coverage profiles (Pearson correlation coefficients of up to 0.99) across multiplexed samples, with <10% strand bias. However, the SNP calling quality was substantially affected by the choice of tools and mapping strategy. With the aim of reducing computational requirements, we compared conventional whole-genome mapping and SNP-calling with a new faster approach: target-region mapping with subsequent ‘read-backmapping’ to the whole genome to reduce the false detection rate. Consequently, we developed a combined mapping pipeline, which includes standard tools (BWA, SAMtools, etc.), and tested it on public HiSeq2000 exome data from the 1000 Genomes Project. Our pipeline saved 12 hours of run time per Hiseq2000 exome sample and detected ~5% more SNPs than the conventional whole genome approach. This suggests that more potential novel SNPs may be discovered using both approaches than with just the conventional approach. Conclusions We recommend applying our general ‘two-step’ mapping approach for more efficient SNP discovery in t NGS. Our study has also shown the benefit of computing inter-sample SNP-concordances and inspecting read alignments in order to attain more confident results.